Evolution of Language

I have recently been reading some articles arguing various positions on functionalist explanations of language versus generativist explanations. These articles (Joan Bybee, Martin Haspelmath, Frederick Newmeyer) have renewed my interest in language evolution in general and in one book in particular. I bought Andrew Carstairs-McCarthy's "The Origins of Complex Language" some time last year but never got around to reading it. I've picked it up again though and found a quote that I think is worth spreading around:

"In most sciences, the choice between rival hypotheses depends on their relative success in predicting the outcomes of controlled experiments; but in sciences relating to the past, the test of a hypothesis is how much of what is currently observed can be deduced from it as a logical consequence. A good hypothesis is one that is both simple and economical in the sense that a relatively large proportion of what is currently observed can be deduced from it, with relatively little reliance on chance or supplementary assumptions. A good hypothesis is also surprising, in the sense that its consequences include facts that do not seem obviously related to one another (Medawar 1967: 125). The more disparate the facts are that a simple hypothesis accurately predicts, the more unlikely it is that these facts are as they are by accident, or for reasons unconnected to the hypothesis."